We use the numerically exact T-matrix method to model light scattering and absorption by aged smoke aerosols at lidar wavelengths ranging from 355 to 1064 nm assuming the aerosols to be smooth spheroids or Chebyshev particles. We show that the unique spectral dependence of the linear depolarization ratio (LDR) and extinction-to-backscatter ratio (or lidar ratio, LR) measured recently for stratospheric Canadian wildfire smoke can be reproduced by a range of model morphologies, a range of spectrally dependent particle refractive indices, and a range of particle sizes. For these particles, the imaginary part of the refractive index is always less than (or close to) 0.035, and the corresponding real part always falls in the range [1.35, 1.65]. The measured spectral LDRs and LRs could be produced by nearly-spherical oblate spheroids or Chebyshev particles whose shapes resemble those of oblate spheroids. Their volume-equivalent effective radii should be large enough (reff = 0.3 μm or greater) to produce the observed enhanced LDRs. Our study demonstrates the usefulness of triple-wavelength LDR measurements as providing additional size information for a more definitive characterization of the particle morphology and composition. Non-zero LDR values indicate the presence of nonspherical aerosols and are highly sensitive to particle shapes and sizes. On the other hand, the LR is a strong function of absorption and is very responsive to changes in the particle refractive index.